On 05/06/2016 09:21 AM, Pavel Hrdina wrote: > On Fri, May 06, 2016 at 08:43:25AM -0400, Cole Robinson wrote: >> On 05/05/2016 12:20 PM, Pavel Hrdina wrote: >>> Signed-off-by: Pavel Hrdina <phrdina@xxxxxxxxxx> >>> --- >>> .../generic-graphics-vnc-listen-attr-only.xml | 28 ++++++++++++++++++++ >>> ...generic-graphics-vnc-listen-element-minimal.xml | 30 ++++++++++++++++++++++ >>> ...ic-graphics-vnc-listen-element-with-address.xml | 30 ++++++++++++++++++++++ >>> .../generic-graphics-vnc-manual-port.xml | 28 ++++++++++++++++++++ >>> .../generic-graphics-vnc-minimal.xml | 28 ++++++++++++++++++++ >>> .../generic-graphics-vnc-socket-listen.xml | 30 ++++++++++++++++++++++ >>> .../generic-graphics-vnc-socket.xml | 28 ++++++++++++++++++++ >>> .../generic-graphics-vnc-listen-attr-only.xml | 30 ++++++++++++++++++++++ >>> ...generic-graphics-vnc-listen-element-minimal.xml | 30 ++++++++++++++++++++++ >>> ...ic-graphics-vnc-listen-element-with-address.xml | 30 ++++++++++++++++++++++ >>> .../generic-graphics-vnc-manual-port.xml | 28 ++++++++++++++++++++ >>> .../generic-graphics-vnc-minimal.xml | 28 ++++++++++++++++++++ >>> .../generic-graphics-vnc-socket-listen.xml | 30 ++++++++++++++++++++++ >>> .../generic-graphics-vnc-socket.xml | 28 ++++++++++++++++++++ >>> tests/genericxml2xmltest.c | 7 +++++ >> >> The generic XML parser doesn't have any restriction on duplicate <graphics >> type='vnc'/>, so these could be combined to possibly one XML file. >> >> I know there really isn't much precedent for that at the moment, but given the >> massive size of our test suite and the fact that much of the time is spent on >> redundant parsing and XML formatting, personally I'd like to see things move >> in that direction. > > The idea itself isn't bad, but I think that it's better to have vnc and spice in > separate XML to make it clear which test failed so you know that it's vnc > graphics or spice graphics directly only from the test name. > Yes, agreed with splitting on VNC vs spice at least. > I've tried this command on my desktop: > > "time ./run tests/.libs/qemuxml2xmltest" > > and considering that there are 837 test cases the output of time is: > > real 0m0.821s > user 0m0.746s > sys 0m0.034s > > So we don't spent that much time while parsing and formatting XML. > Fair enough, but truly the big time sink is domainschematest... - Cole -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list